Application Note Semiconductor Analysis of Trace Metal Impurities in High Purity Hydrochloric Acid Using ICP-QQQ Authors Kazuo Yamanaka and Kazuhiro Sakai Agilent Technologies, Japan Introduction Hydrochloric acid (HCl) is a component of the standard RCA cleaning process used to remove organic and metallic residues and impurities from the surface of silicon wafers used in semiconductor manufacturing. The cleaning steps are performed before high temperature processing steps such as oxidation and chemical vapor deposition (CVD). RCA Standard Clean 2 (SC-2) removes ionic contaminants from the wafer surface. SC-2 follows SC-1, which removes organic residues and particles. SC-2 consists of HCl combined with hydrogen peroxide (H 2 ) and de-ionized water (DIW). Since the cleaning solutions are in direct contact with the silicon wafer surface, ultrahigh purity reagents are required for these solutions.
SEMI standard C27-0708 Tier-C protocol for HCl specifies a maximum contaminant level of 100 ppt for each element (HCl 37.0-38.0 %) [1]. The concentration of industrial grade HCl is usually 20 or 35%, depending on the method of production. The Cl matrix leads to the formation of several polyatomic ions, which cause significant spectral interferences on some key elements. For example, H 2 37 Cl + on 39 K +, 35 Cl 16 O + on 51 V +, 35 Cl 16 OH + on 52 Cr +, 37 Cl 16 O + on 53 Cr +, 35 Cl 37 Cl + on 72 Ge +, 37 Cl 2 + on 74 Ge +, and 40 Ar 35 Cl + on 75 As +. As a result of these polyatomic interferences, it has been difficult to determine these elements at the required levels using conventional single quadrupole ICP-MS (ICP-QMS). Even ICP- QMS instruments fitted with a collision/reaction cell (CRC) or bandpass filter can only offer limited reduction of the spectral interferences arising from the Cl matrix. Consequently, some methods for the analysis of high purity HCl by ICP-QMS have recommended sample pretreatment steps to remove the chloride matrix, which can lead to analyte loss and/or sample contamination. In this study, triple quadrupole ICP-MS (ICP-QQQ) was used to analyze 50 elements in HCl, using MS/MS mode to resolve the polyatomic interferences. All analytes, including the most problematic elements such as K, V, Cr, Ge, and As, could be determined directly in the undiluted HCl with single digit ppt detection limits. Experimental Instrumentation An Agilent 8900 Semiconductor configuration ICP-QQQ was used in this study. The instrument was fitted with a PFA-100 nebulizer, Peltier-cooled quartz spray chamber, quartz torch, platinum-tipped sampling and skimmer cones and s-lens. The nebulizer was operated in self-aspiration mode to minimize the potential for sample contamination from the peristaltic pump tubing. In advanced semiconductor applications, the key requirement is to deliver the absolute lowest possible detection limits (DLs) for each analyte. To achieve this goal, laboratories measuring ultratrace levels of contaminants can use a multi-tune method, where several tuning steps are applied sequentially during the measurement of each solution. This approach allows the tuning conditions to be optimized for the removal of different types of interferences, while maintaining maximum sensitivity for each analyte. In this work, several reaction cell gases (H 2,, and NH 3 ) were used as appropriate for the large number of analytes being measured. He was used as a buffer gas in the NH 3 reaction gas modes. Tuning conditions are shown in Table 1 and other acquisition parameters are shown in Table 2. Table 1. ICP-QQQ operating conditions. Acquisition mode Cool Cool- No gas H 2 NH 3 -soft NH 3 MS/MS RF power (W) 600 1500 Sampling depth (mm) 18.0 8.0 Nebulizer gas (L/min) 0.70 Makeup gas (L/min) 0.90 0.48 Extract 1 (V) -150.0 4.2 4.7 4.5 3.5 Extract 2 (V) -18.0-17.0-250.0-120.0 Omega bias (V) -70.0-140.0-70.0 Omega lens (V) 2.0 10.0 8.0 10.5 4.0 Q1 entrance (V) -15.0-50.0 He flow (ml/min) - 1.0 - - - 1.0 - H 2 flow (ml/min) - - - 7.0 - - - NH 3 flow (ml/min) - 2.0 (20%) - - - 2.0 (20%) flow (ml/min) - - - - 0.45 (30%) - - 0.45 (30%) Axial acceleration (V) 0.0 1.5 0.0 1.0 0.2 1.0 Energy discrimination (V) 15.0-5.0 5.0 0.0-7.0 Table 2. Acquisition parameters. Parameter Q2 peak pattern Replicates Setting 1 point 3 (spiked solution) 10 (unspiked solution) Sweeps/replicate 10 Integration time Samples and standards 2 s for all isotopes The samples of HCl used in this study included: Sample 1: 20% HCl (high purity grade). Sample 2: 36% HCl (non-high purity grade). Sample 3: 20% HCl (34% high purity grade diluted to 20% with DIW). No further sample preparation was necessary as all samples were introduced directly into the ICP-QQQ. To run undiluted HCl routinely, it is recommended that the large (18 mm) insert Pt cone is fitted. Long-term corrosion of internal ICP-MS components can be minimized by fitting the dry pump option. Calibration and quantification were done using the method of standard additions (MSA). Standard solutions were prepared by spiking a multi-element standard solution (SPEX CertiPrep, 2
NJ, US) into each HCl sample type to give spike levels of 10, 20, 30, and 40 ppt. The MSA calibrations were then automatically converted to external calibrations in the ICP-MS MassHunter data analysis table. This conversion allows other samples of the same type (HCl concentration) to be quantified without requiring separate MSA spike additions into each sample. All solutions were prepared just before analysis. All preparation and analysis was performed in a Class 10,000 clean room. Results and Discussion DLs and BECs In total, 50 elements including all SEMI specification analytes were measured using the 8900 ICP-QQQ operating in multiple tune modes. Data for each mode was combined automatically into a single report for each sample. Detection limits (DLs) and background equivalent concentrations (BECs) in 20% HCl are given in Table 3. Table 3. DLs and BECs in high purity 20% HCl*. Element Cell gas mode Q1 mass Q2 mass DL ng/l BEC ng/l Li Cool-NH 3 7 7 0.032 0.016 Be No gas 9 9 0.022 0.021 B No gas 11 11 0.55 4.1 Na Cool-NH 3 23 23 0.064 0.15 Mg Cool-NH 3 24 24 0.077 0.056 Al Cool-NH 3 27 27 0.20 0.19 P -soft 31 47 1.1 2.6 K Cool-NH 3 39 39 0.087 0.17 Ca Cool-NH 3 40 40 0.44 0.68 Sc -soft 45 61 0.014 0.012 Ti -soft 48 64 0.051 0.074 V NH 3 51 51 0.11 0.19 Cr Cool-NH 3 52 52 0.18 0.12 Mn Cool-NH 3 55 55 0.016 0.006 Fe Cool-NH 3 56 56 0.24 0.27 Co Cool-NH 3 59 59 0.10 0.038 Ni Cool-NH 3 60 60 0.66 0.26 Cu Cool-NH 3 63 63 0.10 0.12 Zn NH 3 66 66 0.14 0.097 Ga NH 3 71 71 0.015 0.026 Ge NH 3 74 107 0.90 3.0 Ge NH 3 74 107 0.32 0.77 As 75 91 1.4 48 As 75 91 0.73 6.2 Se H 2 78 78 0.44 0.52 Rb Cool-NH 3 85 85 0.041 0.013 Sr NH 3 88 88 0.003 0.001 Y - soft 90 106 0.010 0.006 Zr -soft 93 125 0.012 0.004 Nb 93 125 0.004 0.005 Mo He 98 98 0.13 0.57 Ru He 101 101 0.016 0.003 Pd He 105 105 0.010 0.001 Ag He 107 107 0.032 0.014 Cd He 114 114 0.090 0.10 In He 115 115 0.035 0.021 Sn He 118 118 0.57 3.3 Sb He 121 121 0.66 1.5 Te H 2 125 125 0.37 0.31 Cs NH 3 133 133 0.008 0.019 Ba NH 3 138 138 0.005 0.005 Hf No gas 178 178 0.005 0.004 Ta He 181 181 0.013 0.010 W No gas 182 182 0.039 0.062 Re No gas 185 185 0.12 0.50 Ir No gas 193 193 0.017 0.012 Au He 197 197 0.027 0.022 Tl No gas 205 205 0.007 0.004 Pb H 2 208 208 0.028 0.023 Bi No gas 209 209 0.024 0.030 Th No gas 232 232 0.017 0.021 U No gas 238 238 0.009 0.005 * Shaded rows for Ge and As indicate results measured in Sample 3, due to suspected contamination for these elements in Sample 1. Quantitative results Table 4 shows quantitative data for all SEMI specification elements in high purity 20% HCl and non-high purity 36% HCl determined by MSA. The results show that the 8900 ICP-QQQ can measure contaminants in HCl at a much lower level than the 100 ppt maximum limit specified in the SEMI specifications. It is important to note that the concentration specified by SEMI is for 37 38% HCl while the data presented here is for 20 and 36% HCl. Even taking this difference into account, the 8900 ICP-QQQ is clearly able to measure contaminants at levels far lower than current industry requirements for high-purity HCl. 3
Table 4. Quantitative results for SEMI specification elements in high purity 20% HCl (Sample 1) and non-high purity 36% HCl (Sample 2). Element Cell gas mode Q1 Q2 Sample 1 20% HCl, ng/l Sample 2 36% HCl, ng/l Li Cool-NH 3 7 7 <DL <DL 0.032 B No gas 11 11 4.1 15 0.55 Na Cool-NH 3 23 23 0.15 6.4 0.064 Mg Cool-NH 3 24 24 <DL 6.5 0.077 Al Cool-NH 3 27 27 <DL 23 0.20 K Cool-NH 3 39 39 0.17 1.5 0.087 Ca Cool-NH 3 40 40 0.68 13 0.44 Ti - soft 48 64 0.074 1.4 0.051 V NH 3 51 51 0.19 4.6 0.11 Cr Cool-NH 3 52 52 <DL 0.55 0.18 Mn Cool-NH 3 55 55 <DL 0.071 0.016 Fe Cool-NH 3 56 56 0.27 7.6 0.24 Ni Cool-NH 3 60 60 <DL <DL 0.66 Cu Cool-NH 3 63 63 0.12 0.57 0.10 Zn NH 3 66 66 <DL 1.1 0.14 As 75 91 48 39 0.73* Cd He 114 114 0.10 0.34 0.090 Sn He 118 118 3.3 2.3 0.57 Sb He 121 121 1.5 0.95 0.66 Ba NH 3 138 138 0.005 <DL 0.005 Pb H 2 208 208 0.023 0.13 0.028 *DL for As measured in Sample 3, due to suspected contamination for this element in Sample 1. DL, ng/l Cr and K determination Cool plasma is a proven technique used to remove plasmabased interferences. Although it has been largely superseded by CRC methodology, cool plasma remains the most effective analytical mode for some elements in certain matrices. Combining cool plasma with CRC technology has been shown to be a powerful mode for interference removal [2]. Because the major isotope of chromium ( 52 Cr + ) suffers an interference from 35 Cl 16 OH + in high purity HCl, Cr was determined using cool plasma with ammonia cell gas. The calibration curve for 52 Cr shows that 35 Cl 16 OH + interference was removed successfully, allowing a BEC of 0.12 ng/l (ppt) to be achieved, with a detection limit of 0.18 ppt (Figure 1). The DL and BEC displayed in the ICP-MS MassHunter calibration plots are based on the 10 replicates of the unspiked high-purity 20% HCl sample. Figure 1. 52 Cr calibration curve obtained using cool plasma and NH 3 cell gas, showing low BEC and good linearity. The same approach is effective for the determination of other interfered elements such as K. Figure 2 shows that the interference from H 2 37 Cl + on 39 K + was suppressed using cool plasma and NH 3 cell gas, giving a BEC and DL for K of 0.17 ppt and 0.09 ppt, respectively. 4
The ClCl + interference on 74 Ge was avoided by measuring a Ge-ammonia cluster ion, 74 Ge[ 14 NH 2 ( 14 NH 3 )] +, in mass-shift mode at mass 107. Q1 (set to m/z 74 to allow the 74 Ge + precursor ions to enter the cell) rejects all non-target masses, including 107 Ag +, which would otherwise overlap the Ge- NH 3 product ion mass. Q1 (in contrast to a bandpass filter) also rejects all other nearby analyte ions, 70 Zn +, 71 Ga +, 73 Ga +, 75 As +, 78 Se +, etc., preventing them from forming potentially overlapping ammonia clusters at the target product ion mass. Representative calibration curves for V and Ge are shown in Figure 3, again illustrating the low BEC (0.19 ppt for V and 0.77 ppt for Ge) and DL (0.11 ppt for V and 0.32 ppt for Ge) achieved with the 8900 with NH 3 cell gas in MS/MS mode. Figure 2. 39 K calibration curve obtained using cool plasma and NH 3 cell gas. V and Ge determination ICP-QMS fitted with a CRC operating in helium collision mode can successfully eliminate many polyatomic ions using He collision cell gas and kinetic energy discrimination (KED) [3]. However, ICP-QMS has some serious limitations when highly reactive cell gases, such as NH 3, are used in the CRC. ICP-QMS has no mass selection step before the cell, so all ions enter the CRC. It is likely, therefore, that new reaction product ions will form in the CRC that may overlap the target analyte mass of interest. Bandpass ICP-QMS instruments, where all ions within a certain mass range (usually about 10 u) of the target analyte can enter the cell and react, have similar limitations to traditional ICP-QMS in terms of controlling reaction chemistry with highly reactive cell gases. ICP-QQQ with MS/MS removes this limitation, as the first quadrupole mass filter (Q1), which is located before the CRC, allows precise selection of the specific mass of ions that are allowed to enter the cell. This extra mass selection step ensures that reaction processes in the cell are controlled, which removes the potential for non-target product ion overlaps and dramatically improves the detectability of the analyte ions. MS/MS acquisition mode using NH 3 as the reaction cell gas was used for the trace determination of V and Ge. The ClO + interference on 51 V was removed using NH 3 on-mass mode. Potentially, 14 NH 2 35 Cl + could form in the cell and interfere with V at m/z 51. However, the unit mass resolution of Q1 on the 8900 ICP-QQQ ensures that only ions at m/z 51 can enter the cell. All other matrix and analyte ions, e.g. 35 Cl +, are prevented from entering the cell and cannot, therefore, contribute to the signal at the analyte mass. This simple approach avoids the formation of any new product ion interferences on 51 V. Figure 3. 51 V and 74 Ge calibration curve obtained using NH 3 cell gas. Determination of As Arsenic has a single isotope at m/z 75 that suffers an interference from the polyatomic ion 40 Ar 35 Cl +. Since ArCl + readily forms in a chloride matrix, the polyatomic interference compromises the determination of As at ultratrace levels in 5
concentrated HCl using ICP-QMS. Oxygen can be used as the cell gas to avoid this overlap, with As being measured as the AsO + product ion at m/z 91. However, with ICP-QMS, the AsO + product ion at mass 91 suffers an interference from 91 Zr +. Helium collision mode in the Agilent ORS cell can reduce ArCl + effectively, allowing a BEC of less than 20 ppt to be achieved by ICP-QMS [3]. But, as semiconductor industry demands become more stringent, this sensitivity may not be sufficient for the lowest level of ultratrace analysis. Using the 8900 ICP-QQQ with MS/MS, the 91 Zr + ion is removed by Q1, which is set to the As + precursor ion mass of 75. MS/MS mode allows cell gas to be used successfully, with As being measured as the AsO + product ion at m/z 91 without overlap from 91 Zr +. A further benefit of cell gas is that measuring AsO + provides more sensitivity that direct measurement of As + in He mode. A calibration curve for As in 20% HCl (Sample 3) is shown in Figure 4, demonstrating a BEC of 6.17 ppt and a DL of 0.73 ppt. While lower than the industry requirements for highpurity HCl, this BEC doesn t represent the best performance that can be achieved with the 8900 ICP-QQQ, so further investigation was done to identify the cause of the relatively high background. AsO + at m/z 91. This possibility can be checked by comparing the isotopic signature of the Cl-based product ions observed in the mass spectrum. Since chlorine has two isotopes, 35 and 37, the ratio of the natural abundances of these isotopes (75.78%: 24.22%) can be used to confirm whether a product ion is Cl-based. The signals of the mass-pairs 75/91 and 77/93, representing the potential Cl interferences 40 Ar 35 Cl 16 O + and 40 Ar 37 Cl 16 O + respectively, were measured by ICP-QQQ with MS/MS. A neutral gain scan spectrum (where Q1 and Q2 are scanned synchronously, with a fixed mass difference between them) was measured and the scan is presented in Figure 5. For this neutral gain scan, Q1 was scanned across the mass range from 74 to 78 u to pass any precursor ions to the CRC, and Q2 was scanned synchronously at Q1 + 16, monitoring any product ions formed by O-atom addition. The peak at masspair m/z 75/91 that caused the relatively high BEC for As in Sample 1 is clearly visible. However, if the signal at 75/91 was due to interference from 40 Ar 35 Cl 16 O +, there would also be a corresponding signal from 40 Ar 37 Cl 16 O + at mass-pair 77/93. Since there was no signal observed at 77/93, we can conclude that the signal at m/z 75/91 is not due to any contribution from ArClO +, and the high reported concentration of As in Sample 1 is due to contamination. Figure 4. 75 As MSA calibration curve obtained in Sample 1 using cell gas. Investigation of arsenic contamination As the measured result for As was relatively high in high purity HCl Sample 1 (Table 4), the signal count at m/z 91 (mass of the product ion AsO + ) was investigated further. In a high Cl matrix, the polyatomic ion 40 Ar 35 Cl + forms in the plasma and during ion extraction. This polyatomic ion has the same nominal mass as the target 75 As + precursor ion, so it passes through Q1 and enters the cell. While not thermodynamically favored, the ArCl + might react with the cell gas to form ArClO +, which would therefore remain as an interference on Figure 5. Neutral gain scan spectrum for 20% high purity HCl showing the theoretical isotope template for 40 Ar 35 Cl 16 O + and 40 Ar 37 Cl 16 O +. Q1 was scanned from m/z 74 to 78, while Q2 was set to Q1 + 16. 6
Conclusions The high performance of Agilent ICP-QQQ systems for the analysis of trace metallic impurities in concentrated HCl has been described previously [4]. Now, the Agilent 8900 Semiconductor configuration ICP-QQQ with flexible cell gas support, unique MS/MS capability, and unparalleled cool plasma performance, further improves the detection limits for the analysis of a wide range of trace metal contaminants in high purity acids. The advanced reaction cell methodology supported by the 8900 ICP-QQQ allows the SEMI elements, including those elements with potential matrix-based interferences such as K, V, Cr, Ge, and As, to be determined at lower concentrations in a chloride matrix than was previously possible. References 1. SEMI C27-0708, Specifications and guidelines for hydrochloric acid (2008) 2. Junichi Takahashi and Katsuo Mizobuchi, Use of Collision Reaction Cell under Cool Plasma Conditions in ICP-MS, 2008 Asia Pacific Winter Conference on Plasma Spectroscopy 3. Junichi Takahashi, Direct analysis of trace metallic impurities in high purity hydrochloric acid by Agilent 7700s/7900 ICP-MS, Agilent publication, 2017, 5990-7354EN 4. Handbook of ICP-QQQ Applications using the Agilent 8800 and 8900, Agilent publication, 2017, pp 13-14, 5991-2802EN More Information When analyzing 20 36% HCl on a routine basis, it is recommended to use the following options: G3280-67056 G4915A G3666-67030 Pt sampling cone (18 mm insert) Upgrade to dry pump Interface valve kit - ball type valve Since hydrochloric acid is corrosive, avoid placing open sample bottles near the instrument. 7
www.agilent.com/chem This information is subject to change without notice. Agilent Technologies, Inc. 2017 Printed in the USA, November 9, 2017 5991-8675EN